University of Strathclyde - Issues and Challenges in Sedentary … · 2020. 9. 30. · Total time spent in sedentary behavior (T) can be derived by summing time in sedentary bouts,

Issues and Challenges in Sedentary Behavior

Measurement

Minsoo Kang

Department of Health and Human Performance, Middle Tennessee State University, Murfreesboro,

5 Tennessee

David A. Rowe School of Psychological Sciences and Health, University of Strathclyde, Glasgow, UK

Measurement in Physical Education and Exercise Science, 00: 1–11, 2015 ISSN: 1091-367X print / 1532-7841 online

DOI: 10.1080/1091367X.2015.1055566

For Peer Review O

nly

Issues and Challenges 2

2

Abstract

Previous research has shown the negative impact of sedentary behavior on health,

including cardiovascular risk factors, chronic disease-related morbidity, and mortality.

Accurate measurement of sedentary behavior is thus important to plan effective interventions

and to inform public health messages. This paper a) provides an overview of the nature and

importance of sedentary behavior, b) describes measurement methods, including subjective

and objective measurement tools, c) reviews the most important measurement and data

processing issues and challenges facing sedentary behavior researchers, and d) presents key

findings from the most recent sedentary behavior measurement-related research. Both

subjective and objective measures of sedentary behavior have limitations for obtaining

accurate sedentary behavior measurements compliant with the current definitions of

sedentary behavior, especially when investigating sedentary behavior as part of the full

spectrum of physical behaviors. Regardless of the sedentary behavior measure chosen,

researchers must be aware of all possible sources of error inherent to each technique and

minimize those errors thereby increasing validity of the outcome data.

Keywords: validity, objective measure, subjective measure, accelerometer, public health,

physical activity

Page 2 of 36

URL: http://mc.manuscriptcentral.com/hmpe Email: [email protected]

Measurement in Physical Education and Exercise Science

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


3

The chronology of population health recommendations associated with human

movement could be generally characterized as (a) do planned, structured exercise in order to

improve fitness (American College of Sports Medicine, 1978); (b) incorporate lifestyle

activities of moderate to vigorous intensity on 30 minutes per day (MVPA; Pate et al., 1995);

and (c) accumulate lifestyle activities of moderate to vigorous intensity for 150 minutes per

week (US Physical Activity Guidelines Committee, 2008). More recently, population public

health recommendations have emerged for sedentary behavior in Canada (Canadian Society

for Exercise Physiology, 2012; Tremblay et al., 2011) and Australia (Australian Government

Department of Health, 2014; Brown, Bauman, Bull, & Burton, 2012).

Sedentary behavior is a major health risk, independent of being physically inactive

(i.e., obtaining an insufficient amount of MVPA). For example, sedentary behavior has been

demonstrated to be an independent risk factor for poor health and premature death (Tremblay,

Colley, Saunders, Healy, & Owen, 2010). Even for those who meet the public health

recommendations for physical activity, too much sitting can compromise metabolic health

(Healy, Dunstan, Salmon, Shaw, et al., 2008; Salmon et al., 2011). Thus, the empirical

evidence supports the conceptual portrayal of sedentary behavior (too much sitting with low

energy expenditure) as a separate and independent construct from inactivity (doing

insufficient health-enhancing physical activity). Considering that sitting is the most common

form of sedentary behavior and that people spend about 55% of their time in sedentary

behavior during the day (Matthews et al., 2008), reducing sitting time has become an

important public health strategy for chronic disease prevention.

Accurate measurement of sedentary behavior is critical to determining the relationship

between sedentary behavior and health, to planning effective interventions, and to informing

public health messages. As population life expectancy increases, it is crucial to extend current

Page 3 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


4

knowledge on the appropriate measurement of sedentary behavior in health outcome research

(Rosenberger, 2012). Continued efforts are being made to reduce measurement error in

assessing sedentary behavior, but due to the changing definitions of sedentary behavior and

the challenges of capturing its two primary components (posture and energy expenditure),

many critical measurement challenges remain unresolved.

The purpose of this article is to provide an overview of the nature and importance of

sedentary behavior, to describe the primary measurement methods including subjective and

objective measurement tools, to review important sedentary behavior measurement and data

processing issues and challenges, and to present key findings from the most recent sedentary

behavior measurement-related research. It concludes with a list of recommendations to be

considered when measuring sedentary behavior.

What is sedentary behavior?

The definition of sedentary behavior has evolved considerably in the past 10 years.

Interestingly, its roots lie in the Latin sedere, meaning “to sit”. In early physical activity

recommendations and physical activity epidemiology research, the term “sedentary” (or

sometimes “essentially sedentary”) was synonymous with “inactive/low active” (e.g.,

Paffenbarger, Hyde, Wing, & Hsieh, 1986), or “inactive/irregularly active” (e.g., Centers for

Disease Control and Prevention, 1993). Predominant current thinking is that inactivity and

sedentariness are separate constructs. Thus, Owen, Healy, Matthews, and Dunstan (2010)

proposed sedentariness to be defined that “sedentariness (too much sitting) is distinct from

too little exercise” (p. 105). More specifically, this newer conceptualization has been

characterized as prolonged sitting, requiring low levels of energy expenditure ranging from

1.0 to 1.5 metabolic equivalent units (METs) (Owen, Bauman, & Brown, 2009; Pate, O'Neill,

& Lobelo, 2008). In 2012, the Sedentary Behavior Research Network published an updated

Page 4 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


5

definition of sedentary behavior as “any waking behavior characterized by energy

expenditure ≤ 1.5 METs while in a sitting or reclining posture” (p. 540). From these sources,

there appears to be a broad contemporary consensus that sedentariness involves both a

postural aspect (sitting or lying) and low levels of energy expenditure, and does not include

light activity (e.g., quiet standing).

Energy expenditure or the amount of movement is relatively homogenous during

sedentary behavior whereas physical activity has several intensity categories and movement

patterns (such as upper body, whole-body, ambulatory, and stationary). In some aspects the

physical mode of sedentary behavior also is far more homogenous (sitting or lying),

compared to physical activity. Perhaps the more challenging aspect of measuring sedentary

behavior is that its temporal pattern is far more complex, because it occurs throughout the day

and is broken up into multiple bouts of varying length making the temporal patterning aspect

of sedentary behavior the most difficult to interpret.

From a pure movement or energy expenditure perspective, Tremblay et al. (2010)

conceptualized sedentary behavior as being at one end of a physiological continuum, above

sleep, with vigorous intensity physical activity at the opposite end of the continuum.

Tremblay et al. (2010) also described the components of sedentary behavior using the

acronym SITT (Sedentary bout frequency, Interruptions of sedentary behavior, Time spent in

sedentary behavior, and Type of behavior engaged in while being sedentary). Although not

directly analogous to the FITT (Frequency, Intensity, Time, and Type) dimensions associated

with health-enhancing physical activity, it serves as a useful heuristic and reminder

throughout this paper of the importance of matching conceptualization of the construct to the

methods used to measure it.

Overview of Sedentary Behavior Measurement Methods

Page 5 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


6

Sedentary behavior has been evaluated using a variety of subjective measures

(questionnaire, interview, and activity-recall instruments) and objective measures

(accelerometry-based motion sensors and inclinometers). Several current approaches to

measuring sedentary behavior match dimensions represented by the SITT acronym (Tremblay

et al., 2010). Sedentary bout frequency (S) can be derived from any objective measure of

sedentary behavior and sedentary breaks. Similarly, interruptions (I) can be operationalized

either through sit-to-stand transitions from instruments such as the activPAL, or by detecting

any drop in activity counts below a specific threshold (e.g., 100 counts/minute [cpm] in the

ActiGraph). Total time spent in sedentary behavior (T) can be derived by summing time in

sedentary bouts, via procedures just described, or via questionnaires. For example, the

Occupational Sitting and Physical Activity Questionnaire (Chau et al., 2012) uses an

algorithm to multiply self-reported percent occupational time spent sitting by self-reported

total time spent at work. Lastly, type of behavior engaged in while being sedentary (T) cannot

typically be derived from objective measures of sedentary behavior, but is conveniently

assessed via self-report. More recently, wearable cameras and photographic images have been

used to classify contextual information about specific behaviors while sitting (e.g., Chastin,

Schwartz, & Skelton, 2013; Kerr et al., 2013; Kim, Barry, & Kang, 2015), providing a

relatively nonintrusive method of directly observing behavior type and context.

Subjective Measures of Sedentary Behavior

A common approach to evaluating sedentary behavior is to ask a single question (e.g.,

time spent TV viewing) in an interview or activity-based questionnaire format concerning the

amount of total time spent sitting or lying down (e.g., Clemes, David, Zhao, Han, & Brown,

2012). Because this approach provides only general information regarding the sedentary

lifestyle of an individual, it may not be a complete representation of sedentary behavior and

Page 6 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


7

thus, it can be challenging for researchers or health professionals to develop targeted behavior

change intervention programs to reduce sedentary behaviors based on this type of evidence.

Put simply, basic summary measures provide insufficient data to inform strategies for

intervening in a complex set of behaviors.

Researchers have recently developed measurement tools that assess multiple

sedentary behaviors (e.g., TV viewing, reading, screen time) and domain-specific sedentary

behaviors (e.g., sitting at work or at home, transportation) (Chau, van der Ploeg, Dunn, Kurko,

& Bauman, 2012; Marshall, Miller, Burton, & Brown, 2010; Rosenberg et al., 2010). The

Sedentary Behavior Questionnaire (Rosenberg et al., 2010) and Marshall Sitting

Questionnaire (Marshall et al., 2010) assess time spent being sedentary on weekdays and

weekend days. The Sedentary Behavior Questionnaire is designed to assess the amount of

time spent in nine behaviors (e.g., watching TV, sitting and talking on the phone,

driving/riding in a car, bus or train), and the Marshall Sitting Questionnaire consists of

reports of time spent sitting in five domains (e.g., while at work, while using a computer at

home). Because the tools describe patterns of sedentary behavior throughout an entire week

including the weekend, individualized and targeted interventions can more effectively target

time spent in sedentary pursuits. The Occupational Sitting and Physical Activity

Questionnaire (Chau et al., 2012) is specific to the workplace environment and asks

participants to report how many hours they worked in the previous 7 days and the number of

days they were at work. It subsequently asks for percent time at work spent sitting, standing

and in physical activity. A basic algorithm is then used to compute total sitting time at work

during one week.

Subjective measures have historically been preferred in large-sample observational

studies of physical activity due to relatively low administration costs and participant burden

Page 7 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


8

(Sallis & Saelens, 2000), and the same has recently been true for use of surveys for large-

sample measurement of sedentary behavior (Atkin et al., 2012). Subjective measures are also

useful to identify the type of behavior and the context in which the sedentary behavior occurs,

which is typically not possible with objective measurement methods. The reported validity

coefficients of subjective measures, however, have varied greatly among studies (range -.02

to .83; Healy, Clark, et al., 2011). Depending on the study, estimates from subjective

measures may over- or underestimate time spent in sedentary behavior (Clark et al., 2011;

Healy, Clark, et al., 2011). Many factors may lead to the inconsistency, including

inappropriate criterion measures used (e.g., using motion sensors instead of using direct

observation), different qualitative attributes (e.g., recall period and question/response format),

mode of administration (e.g., interview vs. self-administration), the time frame of assessment

(e.g., past day, past week, usual week, past year), population being assessed (e.g., children,

adults and older adults), and cultural norm and social desirability of the response (Atkin et al.,

2012; Healy, Clark, et al., 2011). Furthermore, sedentary behavior is not commonly

structured and purposive like physical activity; rather, it occurs persistently throughout the

day. This may negatively impact participants’ ability to recall accurately the amount of time

spent in sedentary behaviors in free-living environments (Healy, Clark, et al., 2011; Owen et

al., 2010).

To overcome some of these recall-related problems, diaries and ecological momentary

assessment (EMA) were developed to record patterns (i.e., the temporal combination of

activities) of sedentary behaviors. EMA is a strategy that can simultaneously capture a

behavior and the factors that may influence it by allowing participants to report their current

activity, location, and social surroundings (Dunton, Liao, Intille, Spruijt-Metz, & Pentz,

2011). One of the major advantages is its ability to provide ecologically valid (“real-world”)

Page 8 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


9

information on an individual’s sedentary behavior patterns. Other advantages include the

small expense in data collection and the ability to administer to many participants at the same

time.

Limitations of the diary and EMA method are the relatively heavy participant burden,

since they both require a high level of adherence to reporting protocols, and the potential for

participant reactivity (Atkin et al., 2012). Reactivity (sometimes referred to as the Hawthorne

Effect; Campbell & Stanley, 1966) is the phenomenon of altering natural behavior patterns

purely as a result of measuring the behavior. Thus, regular self-monitoring may increase a

participant’s awareness of their sedentary behavior, which in turn may cause them to alter

their behavior by, for example, breaking up their patterns of sitting more frequently than if

they were not self-monitoring. While this may be a positive outcome within “measurement as

intervention” type studies (such as the use of self-monitoring for promoting behavior change),

it is undesirable in contexts where the primary goal is to accurately capture information on

habitual sedentary behavior.

Objective Measures of Sedentary Behavior

Technological advancements in recent years have led to increased accessibility to, and

use of, objective measurement of sedentary behavior. Objective measures of sedentary

behavior can be categorized into two types: energy expenditure devices and posture

classification devices (Granat, 2012). Importantly, most of these devices use the same

underlying technology (accelerometry) but use different algorithms to interpret the data to

estimate either energy expenditure or body position. Energy expenditure devices, which

generally comprise accelerometers such as the ActiGraph GT3X (ActiGraph, Pensacola, FL),

are typically worn on the waist or wrist. They measure human movement and provide the

magnitude of the acceleration within set time periods (epochs), or as a continuous data stream

Page 9 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


10

at frequencies such as 50 Hz. This is then translated into proprietary activity counts (i.e., the

underlying technology is the collection of streaming acceleration data and subsequent

conversion into counts in time epochs using proprietary algorithms and software). The

thresholds of activity counts that correspond to the energy expenditure levels associated with

sedentary behavior (≤ 1.5 METs) vary somewhat, but a threshold of less than 100 cpm has

generally been used with vertical axis data from the ActiGraph (Matthews et al., 2008). As

such devices are now capable of measuring acceleration in three axes, research is ongoing to

develop equivalent cutpoints using the three-axis vector magnitude.

Posture classification devices, such as the activPAL (PAL Technologies Ltd, Glasgow,

UK), provide data on absolute body positions or status of human movement. Similar to

energy expenditure devices, the derived data (e.g., incline of the thigh) are continuous and

based on body segment acceleration, and decision criteria are used to decide whether to

classify incline of the instrument as horizontal (sitting/lying) or vertical (standing). The

activPAL device is capable of monitoring shifts in posture (e.g., moving from sitting or lying

to standing and vice versa) by assessing motion in the vertical, horizontal, and anterio-

posterior planes, functions as an inclinometer, and provides outputs on three types of postural

activities (i.e., sitting/lying, standing, and stepping) using an event file (Grant, Ryan, Tigbe,

& Granat, 2006). Thus, time stamped data record when the wearer transitions from one mode

(sit/lie, stand or step) to another. The exact angle of the thigh that corresponds to transitioning

from a sitting to standing position, and that corresponds to transitioning from standing to

sitting, are different, as applied by the proprietary algorithms. Specifically, a differential

threshold is applied, where the angle determining a sit-to-stand transition is closer to vertical

than the angle used to determine a stand-to-sit transition. Manufacturer software is then used

to reduce data from the event file into summary data on time spent sitting/lying, standing, and

Page 10 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


11

walking, total steps, and other outcome variables that are facilitated by the time-stamped

event recordings such as number of bouts in each mode, average length of each bout for each

mode, number of transitions from sedentary to upright position, and time patterning (time of

day associated with different behavior patterns).

Regardless of whether a primarily energy-expenditure-driven instrument or a

primarily posture-driven instrument is used, similar processes underlie the generation of data

from objective devices, as represented in Table 1. Error can be introduced at any stage of the

process, from the measurement of raw acceleration in the initial stages, through the

application of firmware and software algorithms for preliminary processing, to the decision

rules applied in the latter stages. The latter includes such decisions as what constitutes a valid

day of wear, what is the minimum required days to constitute reliable and valid data, and

minimum bout length criteria. From a strict measurement perspective, errors at only the first

stage of the process could be labelled as measurement error. However, because acceleration is

not the primary construct of interest, we propose that any of the subsequent stages of the

process should be considered as possible sources of error. As we show later in this paper,

decisions such as what cutpoints to apply, definitions of minimal bout lengths, and

parameters for defining sedentary breaks all can influence the variability in outcome variables

representing the various dimensions of sedentary behavior represented by the SITT acronym.

Similar to many other areas of kinesiology such as the prediction of maximal aerobic fitness

or the prediction of body composition components, all sources of variability that are not

attributable to true score variability should be considered, including aspects such as data

reduction techniques, application of software algorithms and rules associated with valid wear

time.

Objective measures of sedentary behaviors are increasingly used as they are believed

Page 11 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


12

to provide more valid and reliable estimates of total time spent in sedentary behaviors in a

free-living environment compared to subjective measures of sedentary behaviors (Atkin et al.,

2012; Healy, Clark, et al., 2011). They also provide more detailed data on temporal pattern.

There are, however, several challenges of using objective measures.

First, both types of sedentary behavior monitoring devices (accelerometry-based

motion sensors and posture sensors) have limited functional abilities for measuring sedentary

behavior in accordance with the prevailing conceptual definition of sedentary behavior. This

may lead to biased estimates of time spent in sedentary behaviors in a free-living

environment. For instance, some accelerometers may mis-classify as sedentary behavior

activities such as quiet standing or some light-intensity activities in a standing position where

the activity counts for those activities are below 100 per minute (Granat, 2012; Marshall &

Merchant, 2013). Conversely, activPAL does not directly implement the definition regarding

energy expenditure (≤1.5 METs) in measuring sedentary behaviors and would classify some

active sitting or lying activities with high energy expenditure (>1.5 METs) such as weight

lifting as sedentary behaviors.

Notably, much health-related research will combine the investigation of sedentary

behaviors and physical activity behaviors. The conceptual representation of sedentary

behaviors as part of a spectrum of physical behaviors including sleep, sedentary behavior and

physical activity, as previously described by Tremblay et al. will probably become standard

within future public health research. This theoretical perspective is taking hold via

organizations such as the International Society for the Measurement of Physical Behaviour

(see http://www.ismpb.org/) and disseminated bi-annually at the International Congress of

Ambulatory Monitoring of Physical Activity and Movement. In terms of objective monitoring

of the spectrum of physical behaviors, it is probably fair to say that the activPAL (which is

Page 12 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


13

unique in currently being the only such commercially-available device) is more accurate at

measuring the sedentary end of the spectrum than the physical activity end of the spectrum

(because currently, physical activity data from the activPAL is limited to step counts and

stepping frequency, or cadence). Conversely, waist-worn accelerometers are more accurate at

measuring the physical activity end of the spectrum than the sedentary end (because they do

not accurately measure posture).

However, recent developments reveal that both instruments, in association with

advances in associated software and algorithms, are improving the accuracy of measuring a

variety of physical behaviors. For example, the ActiGraph currently incorporates an

inclinometer for use when worn at the waist. Carr & Mahar evaluated its accuracy for

classifying inactive lying and sitting and inactive standing, and found classification accuracy

to be only 61%-67%. Interesting research has been conducted recently on the use of wrist-

worn accelerometry for identifying sedentary behaviors. Rowlands et al. (2014) investigated

the classification accuracy of the GENEActive triaxial accelerometer for determining posture,

using the activPAL as the criterion measure. In three small but diverse samples, classification

accuracy from the wrist-worn GENEActive for sitting/lying and standing was generally

moderate to high, as evidenced by inter-instrument correlations, kappa, and mean differences.

Some of the results were not as positive, but this method is still at the developmental stage

and shows promise for measuring the broader spectrum of physical behaviors.

Second, the appropriateness of the activity counts threshold between sedentary

behavior and light-intensity activity is of particular importance (Kim, Lee, Peters, Gaesser, &

Welk, 2014; Kozey-Keadle, Libertine, Lyden, Staudenmayer, & Freedson, 2011). The activity

count threshold of 100 cpm for ActiGraph single-axis data is generally accepted in adults, but

the validity evidence and justification of this threshold is quite limited (Atkin et al., 2012;

Page 13 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


14

Matthews et al., 2008). Kozey-Keadle et al. (2011) evaluated different thresholds of activity

counts (50, 100, 150, 200, and 250 cpm) in a sample of 20 overweight, inactive office

workers. They found that compared to direct observation, the ActiGraph GT3X 100 cpm

threshold underestimated sitting time by approximately 5% while the 150 cpm threshold

demonstrated the lowest percent bias (1.8%). Kim et al. (2014) evaluated the criterion-

referenced validity of several sedentary thresholds (100, 200, 300, 500, 800, and 1,100 cpm)

using the ActiGraph GT3X placed on the hip and wrist for children while performing a series

of prescribed activities ranging from sedentary to vigorous. Criterion classification was based

on compendium MET values and each activity was dichotomized as either sedentary or non-

sedentary. Results supported the use of a

For Peer Review O

nly


15

Third, wear time can significantly influence estimation of sedentary behavior. This is

likely to be greater for sedentary behavior than for physical activity, as sedentary behavior

comprises more daily time than total physical activity time, and also than time spent in any

physical activity intensity (light, moderate, vigorous, etc.). While it is commonly accepted

that the minimum wear time for accelerometer derived physical activity is ≥10 hours per day

(h/day), there are no guidelines for minimum wear time for sedentary behavior. Herrmann,

Barreira, Kang and Ainsworth (2013) used a semi-simulation approach in 124 adults

participating in a workplace walking intervention to investigate the effect of wear time on

various physical activity intensities and on sedentary time. They found that time spent in

sedentary behavior (

For Peer Review O

nly


16

with a minimum wear-time of 12 hours was 18% (7-day minimum), 50% (5-day minimum),

and 72% (3-day minimum) compared to compliance values for 10 hours of 33% (7-day

minimum), 67% (5-day minimum), and 83% (3-day minimum). Overall, this corresponded to

data loss of 11%-17% for a 12-hr minimum wear time compared to a 10-hr minimum wear

time.

One solution may be to individually impute data from shorter wear times to a defined

“maximum day”. For example, if the maximum day were defined as 16 hours of wake time,

and sedentary time from 10 hours of wear was determined to be 8.5 hours (i.e., 85% of wear

time), imputed sedentary time would be 13.6 hours (85% of 16 hours). Such calculations are

based on an assumption of “equal qualities” in the missing data - in other words, an

assumption that sedentary behavior during non-wear time was similar to that during wear

time. There is some evidence to support this assumption. At the sample level, in the

Herrmann et al. (2014) study, percent time spent being sedentary was 54.9% for a 14-hr

minimum wear time, compared to 54.5% for 13 hr, 54.3% for 12 hr, 54.2% for 11 hr, and

54.0% for 10 hr. This does not necessarily translate to the level of the individual participant,

and additional semi-simulation research is needed to assess the validity of such imputation.

Recent Developments in Sedentary Behavior Measurement-Related Research

Despite increased awareness of the importance of sedentary behavior, little attention

has been paid to validity of the bout duration of sedentary behavior measurement. Although

sedentary behavior is defined as prolonged sitting or reclining (Owen et al., 2009), the

majority of studies that examined objectively measured sedentary behavior using

accelerometers have used a 1-minute bout as the minimum length of duration for valid

sedentary minutes (Clark et al., 2011; Healy, Matthews, Dunstan, Winkler, & Owen, 2011;

Maher, Mire, Harrington, Staiano, & Katzmarzyk, 2013). This may be a legacy effect, as

Page 16 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


17

most such studies have used NHANES data, which was based on early ActiGraph

accelerometers that were initialized to collect data in 1-min epochs. Aggregating data from 1-

min epochs to estimate total time spent in sedentary behavior from accelerometer data may

lead to inclusion of significant time spent in ‘sporadic sedentary behavior’ that is not

necessarily ‘prolonged sedentary behavior’. It can mask the “interruptions” component of

sedentary behavior – a single 30-minute bout of sitting is treated the same as 15 separate 2-

minute bouts of sitting. The operationalization of sedentary time from 1-min epoch data may

therefore not be congruent with what has been defined at the conceptual level (i.e., from a

health perspective, prolonged periods of continuous sitting are more harmful to health). This

is likely an issue for all types of objective measures of sedentary behavior (motion sensors

and posture sensors) and should be considered very carefully when deciding on data

reduction procedures.

In a study to examine patterns of sedentary behavior among US adults in 2003-2006

NHANES, Kim, Welk, Braun, and Kang (2015) found that the current algorithm to identify

sedentary behavior (a minimum of 1-min bout for sedentary time) in accelerometry data may

not be appropriate to obtain valid measurements of sedentary behavior that are compliant

with the current definition of sedentary behavior. The study examined the influences of

different bout periods (i.e., 1, 2–4, 5–9, 10–14, 15–19, 20–24, 25–29, and 30-min) of

sedentary behaviors on health outcomes, and the results showed that sedentary minutes

obtained by a minimum of 5-min bouts yielded a better model-data fit to predict

cardiovascular risk factors.

Studies have shown that sedentary breaks are positively associated with health

outcomes, independent of total sedentary minutes (Healy, Dunstan, Salmon, Shaw, et al.,

2008; Healy, Matthews, et al., 2011). Healy, Dunstan, Salmon, Cerin, et al. (2008) presented

Page 17 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


18

the idea of two individuals with similar total sedentary time. The “prolonger” accumulated

his time in long, continuous bouts with few interruptions, while the “breaker” accumulated

her time via multiple, shorter bouts with frequent interruptions. The findings of Healy,

Dunstan, Salmon, Shaw, et al. (2008) and Healy, Matthews, et al. (2011) have prompted

many sedentary behavior researchers to modify patterns of sedentary time accumulation by

breaking up prolonged sedentary time as an explicit intervention strategy (Owen et al., 2010).

It is still, however, unclear what a “valid” sedentary break is. In this case, validity does not

refer to accuracy of measurement but appropriateness of the measurement – in order to

determine what is a “valid” break, we need to determine what break characteristics are related

to health benefits. A break in sedentary time is typically operationalized by counting every

transition point from a sedentary to active phase (e.g., from

For Peer Review O

nly


19

(beneficially) associated with health outcomes when delimited to breaks that followed a brief

bout of sedentary behavior, but negatively (detrimentally) associated with health outcomes

when limited to breaks that followed a longer sedentary bout duration. This clearly indicates

that when investigating sedentary breaks, we should not apply the prevailing “one-size-fits-all”

approach in which any postural transition following a period of sedentary behavior is counted

as a break, regardless of how short the sedentary bout was (typically, anything of at least 1

min is currently counted as valid). From the study, it appears that different methods for

computing a sedentary break result in variables that have different implications for health.

The apparent finding that “more frequent” breaks in prolonged sitting are detrimentally

associated with health seems counterintuitive. However, it simply reflects the fact that the

number of breaks following prolonged sedentary bouts serves as a proxy for prolonged sitting

(i.e., a break can only occur if a prolonged bout occurs). Interestingly, the total number of

sedentary breaks was identical or slightly less than the accrued number of sedentary bouts of

at least 1-min bout duration. In other words, counting the number of transition points from a

sedentary to active phase would produce a very similar number compared to the number of

sedentary bouts (Kim et al., 2015). This finding indicates that we might be incorrectly

measuring the conceptual definition of valid sedentary breaks (interruptions in sedentary time)

using the current method of processing data to arrive at sedentary breaks. Further research is

needed to improve the validity of objectively measured sedentary breaks.

Besides temporal issues related to sedentary bout length and sedentary breaks, there

are broader temporal conundrums in the assessment of total daily sedentary time. Unlike

MVPA, which is largely temporally independent of waking time, sedentary time is much

more temporally dependent on wake time. The pie chart in Figure 1 illustrates why. Physical

behavior during total daily time (24 hr) can be broken up into four types of behaviors. A small

Page 19 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


20

time period is generally spent in MVPA. For example, from national accelerometry data in

the US, adults in various age and sex categories spend on average only 5.4 to 42.8 min/day in

MVPA (Troiano et al., 2008), even using very liberal criteria (not applying a restrictive 10-

min bout condition). This corresponds to only 0.4% to 3.0% of each daily 24-hr cycle.

Conversely, mean sleep time in a similar sample of adults ranged from 6.7 to 7.4 hr, or 28%

to 31% of a daily 24-hr cycle (Ram, Seirawan, Kumar, & Clark, 2010). Also using NHANES

accelerometry data, Matthews et al. (2008) determined average time spent by adults in

sedentary behavior (cpm < 100) to be 7.5 hr to 9.3 hr, or 31% to 39% of a daily 24-hr cycle,

adjusted for wear time. These proportions are approximated in Figure 1, from which it is

evident that MVPA comprises a trivial proportion of the 24-hr day, and sleep, light intensity

physical activity and sedentary time comprise approximately one-third each of total daily

time.

Certain mathematical conclusions can be derived from this information. First, changes

in MVPA (and errors in measurement of MVPA) will have no meaningful displacement effect

on time spent in sleep, light intensity physical activity and sedentary behavior. Second, if

sleep is held constant, displacement of sedentary time will be directly related to changes in

light intensity physical activity time. This latter conundrum has been addressed recently from

both a measurement perspective and a behavior modification perspective (i.e., in order to

reduce or break up prolonged sedentary time, behavioral interventions often target increasing

light intensity physical activity). From a measurement or statistical perspective, the effect of

reduction in sedentary time on health, and relationships between sedentary time and health,

are collinearly associated with similar effects (or relationships) between light intensity

physical activity and health. This was demonstrated by Maher, Olds, Mire, and Katzmarzyk

(2014), who found that when adjusted for total physical activity (i.e., the sum of light,

Page 20 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


21

moderate and vigorous intensity physical activity), most correlations between sedentary

behavior and several cardiometabolic markers effectively disappeared. Third, because sleep

(a physical behavior that is positively related to health) varies among individuals and is

directly mathematically related to sedentary time, estimates of sedentary time should be

adjusted to take sleep into account. Pedisic (2014) reviewed 54 studies investigating

relationships between sedentary behavior and health, of which only two studies, investigating

TV time and health, adjusted for sleep time. In his thorough review of the underlying

statistical issues, he proposed the Activity Balance Model, a new theoretical framework for

combining the investigation of sedentary behavior and health with sleep duration, standing

time, light intensity physical activity and MVPA. While there is currently little empirical

evidence to support this framework, it seems inevitable that the measurement and

investigation of sedentary behavior will be inextricably linked with the broader spectrum of

physical behaviors.

Through rapid technological advancements and collaborations across disciplines, new

analytic and modeling approaches to classifying different types of behaviors have been

developed and tested. Lyden, Kozey-Keadle, Staudenmayer, and Freedson (2014) introduced

the sojourn method, a machine learning technique that combines artificial neural networks

with decision tree analysis, and found that two sojourn methods (sojourn-1 Axis and sojourn-

3 Axis) improved the estimation of sedentary time from a single, hip-mounted accelerometer

compared to direct observation over 3 days of free living in 13 adults. More validation studies

regarding the application of analytic and modeling approaches in free-living environments

are needed.

Conclusion

Page 21 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


22

In this paper we have attempted to set the scene for subsequent evidence-based

papers evaluating methods for assessing sedentary behavior. In population-based

observational studies to monitor sedentary behavior, subjective measures have been preferred

for their efficiency and practicality; however, several disadvantages of subjective measures

impede the reliable and valid estimation of sedentary behavior in free-living environments.

Furthermore, while objective measures of physical activity have been shown to provide more

reliable and valid estimates for sedentary behavior compared to subjective measures, they are

still limited in their ability to obtain accurate measurements that comply with the current

definition of sedentary behavior. Specifically, no current widely-available objective tool

accurately detects both posture and energy expenditure, and none accurately measures the full

spectrum of physical behaviors (from sleep to vigorous intensity physical activity).

In this paper, different methods have been described for assessing sedentary behavior,

but no specific method can be identified as the best method for all purposes. Each method has

its own advantages and disadvantages, which must be carefully considered before selecting a

measure. New analytic and modeling approaches for translating raw accelerometer data to

classify different types of sedentary behavior show considerable promise; yet challenges

regarding the application of analytic and modeling approaches in free-living environments

still need to be addressed. Regardless of the sedentary behavior measure chosen, researchers

must be aware of all possible sources of error inherent to each technique and attempt to

minimize those errors thereby increasing the validity of the outcome data.

We present below specific recommendations based on the current evidence and

thinking presented in this paper:

• Subjective measures are primarily suitable for providing general summary

information. They are most suited to use in large-sample descriptive studies,

Page 22 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


23

with the caveat that they provide less reliable estimates of sedentary behavior

than objective measures. They can be used to provide useful information on

context and type of behavior and therefore are a valuable adjunct method for

use with objective measures, in order to understand behavioral aspects and

thereby inform interventions.

• Until technological advances facilitate the combination of both methods in a

single instrument, posture-focused instruments should be used where

sedentary behavior is the primary outcome of interest and energy-expenditure-

focused instruments should be used where differentiating between intensities

of physical activity is the primary goal.

• Wear time influences estimates of sedentary behavior more than for estimating

other physical behaviors. To resolve this problem, either stricter wear time

criteria must be applied or missing data should be imputed to at least reflect a

“standard day”. Further semi-simulation studies are needed in order to

evaluate the effect of various imputation methods on accuracy of full-day

estimates of sedentary behavior.

• Because of the mathematical interdependence of sedentary time and sleep

time, and the fact that sleep time is generally beneficial to health, estimates of

sedentary time should preferably be adjusted to reflect accurately-determined

wake time, possibly assessed via a wake and sleep log.

• When aggregating sedentary time, a minimum bout length of at least 5

continuous minutes should be used from waist-worn accelerometer devices.

Page 23 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


24

• Current methods of computing sedentary breaks incorrectly measure the

conceptual definition of valid sedentary breaks. Thus, caution is needed when

interpreting results about sedentary breaks in relation to health outcomes, and

future research is warranted to improve the validity of objectively measured

sedentary breaks. Parameters relating to the length of the sedentary bout

preceding a break and the length of the break itself also need to be considered.

• Machine-learning techniques show promise for estimating sedentary time from

waist-mounted accelerometry, but as with similar methods for analyzing

physical activity data, we seem to be a long way from widespread

incorporation of these methods into software for processing free-living

acceleration data.

Page 24 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


25

References

American College of Sports Medicine. (1978). ACSM Position statement on the

recommended quantity and quality of exercise for developing and maintaining fitness in

healthy adults. Medicine and Science in Sports and Exercise, 10(3), vi-x.

Atkin, A. J., Gorely, T., Clemes, S. A., Yates, T., Edwardson, C., Brage, S., . . . Biddle, S. J.

H. (2012). Methods of measurement in epidemiology: Sedentary behavior. International

Jounal of Epidemiology, 41, 1460-1471.

Australian Government Department of Health. (2014). Australia’s physical activity and

sedentary behaviour guidelines: Adults. Canberra, Australia: Author.

Brown, W. J., Bauman, A. E., Bull, F. C., & Burton, N. W. (2012). Development of evidence-

based physical activity recommendations for adults (18-64 years): Final report. Canberra,

Australia: Australian Government Department of Health.

Canadian Society for Exercise Physiology. (2012). Canadian physical activity guidelines;

Canadian sedentary behaviour guidelines. Ottawa, Canada: Author.

Campbell, D. T., & Stanley, J. C. (1966). Experimental and quasi-experimental designs for

research. Chicago: Rand McNally.

Carr, L. J., & Mahar, M. T. (2012). Accuracy of intensity and inclinometer output of three

activity monitors for identification of sedentary behavior and light-intensity activity.

Journal of Obesity, 2012, Article ID 460271.

Centers for Disease Control and Prevention. (1993). Prevalence of sedentary lifestyle -

Behavioral Risk Factor Surveillance System, United States, 1991. Morbidity and

Mortality Weekly Report, 42(29), 576-579.

Chastin, S. F. M., Schwarz, U., & Skelton, D. A. (2013). Development of a consensus

taxonomy of sedentary behaviors (SIT): Report of Delphi round 1. PloS ONE, 8(12),

Page 25 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


26

e82313. doi: 10.1371/journal.pone.0082313

Chau, J. Y., van der Ploeg, H. P., Dunn, S., Kurko, J., & Bauman, A. E. (2012). Validity of the

Occupational Sitting and Physical Activity Questionnaire. Medicine and Science in

Sports and Exercise, 44, 118-125.

Clark, B. K., Healy, G. N., Winkler, E. A., Gardiner, P. A., Sugiyama, T., Dunstan, D. W., . . .

Owen, N. (2011). Relationship of television time with accelerometer-derived sedentary

time: NHANES. Medicine and Science in Sports and Exercise, 43, 822-828.

Clemes, S. A., David, B. M., Zhao, Y., Han, X., & Brown, W. (2012). Validity of two self-

report measures of sitting time. Journal of Physical Activity and Health, 9, 533-539.

Dunton, G. F., Liao, Y., Intille, S. S., Spruijt-Metz, D., & Pentz, M. (2011). Investigating

children’s physical activity and sedentary behavior using ecological momentary

assessment with mobile phones. Obesity, 19, 1205-1212.

Granat, M. H. (2012). Event-based analysis of free-living behavior. Physiological

Measurement, 33, 1785-1800.

Grant, P. M., Ryan, C. G., Tigbe, W. W., & Granat, M. H. (2006). The validation of a novel

activity monitor in the measurement of posture and motion during everyday activities.

British Journal of Sports Medicine, 40, 992-997.

Healy, G. N., Clark, B. K., Winkler, E. A. H., Gardiner, P. A., Brown, W. J., & Matthews, C.

E. (2011). Measurement of adults' sedentary time in population-based studies. American

Journal of Preventive Medicine, 41, 216-227.

Healy, G. N., Dunstan, D. W., Salmon, J., Cerin, E., Shaw, J. E., Zimmet, P. Z. , . . . Owen, N.

(2008). Breaks in sedentary time: Beneficial associations with metabolic risk. Diabetes

Care, 31, 661-666. doi:10.2337/dc07-2046

Healy, G. N., Dunstan, D. W., Salmon, J., Shaw, J. E., Zimmet, P. Z., & Owen, N. (2008).

Page 26 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


27

Television time and continuous metabolic risk in physically active adults. Medicine and

Science in Sports and Exercise, 40, 639-645.

Healy, G. N., Matthews, C. E., Dunstan, D. W., Winkler, E. A., & Owen, N. (2011). Sedentary

time and cardio-metabolic biomarkers in US adults: NHANES 2003-06. European Heart

Journal, 32, 590-597.

Herrmann, S. D., Barreira, T. V., Kang, M., & Ainsworth, B. E. (2013). How many hours are

enough? Optimal accelerometer wear time to reflect daily activity. Journal of Physical

Activity and Health, 10, 742-749.

Herrmann, S. D., Barreira, T. V., Kang, M., & Ainsworth, B. E. (2014). Impact of

accelerometer wear time on physical activity data: A NHANES semisimulation data

approach. British Journal of Sports Medicine, 48, 278–282.

Kerr, J., Marshall, S. J., Godbole, S., Chen, J., Legge, A., Doherty, A. R. , . . . Foster, C.

(2013). Using the SenseCam to improve classifications of sedentary behavior in free-

living settings. American Journal of Preventive Medicine, 44, 290-296.

Kim, Y., Barry, V. W., & Kang, M. (2015). Validation of objective monitoring devices for

the assessment of sedentary behavior: Implication of minimum sedentary bout durations.

Measurement in Physical Education and Exercise Science, 19, xxx-xxx.

Kim, Y., Lee, J., Peters, B. P., Gaesser, G. A., & Welk, G. J. (2014). Examination of different

accelerometer cut-points for assessing sedentary behaviors in children. PLoS ONE, 9(4),

e90630. doi:10.1371/journal.pone.0090630

Kim, Y., Welk, G. J., Braun, S. I., & Kang, M. (2015). Extracting objective estimates of

sedentary behavior from accelerometer data: Measurement considerations for

surveillance and research applications. PLoS ONE, 10(2), e0118078.

doi:10.1371/journal.pone.0118078

Page 27 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


28

Kozey-Keadle, S., Libertine, A., Lyden, K., Staudenmayer, J., & Freedson, P. S. (2011).

Validation of wearable monitors for assessing sedentary behavior. Medicine and Science

in Sports and Exercise, 43, 1561-1567.

Lyden, K., Kozey-Keadle, S., Staudenmayer, J., & Freedson, P. S. (2014). A method to

estimate free-living active and sedentary behavior from an accelerometer. Medicine and

Science in Sports and Exercise, 46, 386-397.

Maher, C. A., Mire, E., Harrington, D. M., Staiano, A. E., & Katzmarzyk, P. T. (2013). The

independent and combined associations of physical activity and sedentary behavior with

obesity in adults: NHANES 2003‐06. Obesity, 21, E730-737. doi: 10.1002/oby.20430

Maher, C. A., Olds, T., Mire, E., & Katzmarzyk, P. T. (2014). Reconsidering the sedentary

behavior paradigm. PloS ONE, 9(1), e86403. doi: 10.1371/journal.pone.0086403

Marshall, A. L., Miller, Y. D., Burton, N. W., & Brown, W. J. (2010). Measuring total and

domain-specific sitting: A study of reliability and validity. Medicine and Science in

Sports and Exercise, 42, 1094-1102.

Marshall, S. J., & Merchant, G. (2013). Advancing the science of sedentary behavior

measurement. American Journal of Preventive Medicine, 44, 190-191.

Matthews, C. E., Chen, K. Y., Freedson, P. S., Buchowski, M. S., Beech, B. M., Pate, R.

R. , . . . Troiano, R. P. (2008). Amount of time spent in sedentary behaviors in the United

States, 2003-2004. American Journal of Epidemiology, 167, 875-881.

Owen, N., Bauman, A., & Brown, W. (2009). Too much sitting: A novel and important

predictor of chronic disease risk? British Journal of Sports Medicine, 43, 81-83.

Owen, N., Healy, G, N., Matthews, C. E., & Dunstan, D. W. (2010). Too much sitting: The

population-health science of sedentary behavior. Exercise and Sport Sciences Reviews,

38, 105-113.

Page 28 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


29

Paffenbarger, R. S., Hyde, R., Wing, A. L., & Hsieh, C-C. (1986). Physical activity, all-cause

mortality, and longevity of college alumni. New England Journal of Medicine, 314, 605-

613.

Pate, R. R., O'Neill, J. R., & Lobelo, F. (2008). The evolving definition of "Sedentary".

Exercise and Sport Sciences Reviews, 36, 173-178.

Pate, R. R., Pratt, M., Blair, S. N., Haskell, W. L., Macera, C. A., Bouchard, C. , . . .

Wilmore, J. H. (1995). Physical activity and public health: A recommendation from the

Centers for Disease Control and Prevention and the American College of Sports

Medicine. JAMA, 273, 402-407.

Pedisic, Z. (2014). Measurement issues and poor adjustments for physical activity and sleep

undermine sedentary behaviour research - the focus should shift to the balance between

sleep, sedentary behaviour, standing and activity. Kinesiology, 46, 135-146.

Ram, S., Seirawan, H., Kumar, S. K. S., & Clark, G. T. (2010). Prevalence and impact of

sleep disorders and sleep habits in the United States. Sleep and Breathing, 14, 63–70

Rosenberg, D. E., Norman, G. J., Wagner, N., Patrick, K., Calfas, K. J., & Sallis, J. F. (2010).

Reliability and validity of the Sedentary Behavior Questionnaire (SBQ) for adults.

Journal of Physical Activity and Health, 7, 697-705.

Rosenberger, M. (2012). Sedentary behavior: Target for change, challenge to assess.

International Journal of Obesity, Suppl2, S26-S29. doi: 10.1038/ijosup.2012.7

Rowlands, A. V., Olds, T. S., Hillsdon, M., Pulsford, R., Hurst, T. L., Eston, R. G. , . . .

Langford, J. (2014). Assessing sedentary behavior with the GENEActiv: Introducing the

sedentary sphere. Medicine and Science in Sports and Exercise, 46, 1235-1247.

Sallis, J. F., & Saelens, B. E. (2000). Assessment of physical activity by self-report: Status,

limitations, and future directions. Research Quarterly for Exercise and Sports, 72(2S),

Page 29 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


30

1S-14S.

Salmon, J., Arundell, L., Hume, C., Brown, H., Hesketh, K., Dunstan, D. W. , . . . Crawford,

D. (2011). A cluster-randomized controlled trial to reduce sedentary behavior and

promote physical activity and health of 8-9 year olds: The Transform-Us! Study. BMC

Public Health, 11, 759. doi:10.1186/1471-2458-11-759

Sedentary Behaviour Research Network. (2012). Letter to the Editor: Standardized use of the

terms “sedentary” and “sedentary behaviours”. Applied Physiology, Nutrition, and

Metabolism, 37, 540-542.

Tremblay, M. S., Colley, R. C., Saunders, T. J., Healy, G. N., & Owen, N. (2010).

Physiological and health implications of a sedentary lifestyle. Applied Physiology,

Nutrition, and Metabolism, 35, 725-740.

Tremblay, M. S., Warburton, D. E. R., Janssen, I., Paterson, D. H., Latimer, A. E., Rhodes, R.

E. , . . . Duggan, M. (2011). New Canadian physical activity guidelines. Applied

Physiology, Nutrition and Metabolism, 36, 36-46

Troiano, R. P., Berrigan, D., Dodd, K. W., Mâsse, L. C., Tilert, T., & McDowell, M. (2008).

Physical activity in the United States measured by accelerometer. Medicine and Science

in Sports and Exercise, 40, 181-188.

Troiano, R. P., McClain, J. J., Brychta, R. J., & Chen, K. Y. (2015). Evolution of

accelerometer methods for physical activity research. British Journal of Sports Medicine,

48, 1019-1023.

Tucker, J. (2010). Physical activity levels and cardiovascular disease risk among U.S. adults:

Comparison between self-reported and objectively measured physical activity.

Graduate Theses and Dissertations, Paper 11746.

US Physical Activity Guidelines Committee. (2008). Physical Activity Guidelines Advisory

Page 30 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly


31

Committee report, 2008. Washington: DHSS.

Page 31 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review Only

Table 1. Sources of error at three stages of the data collection and interpretation process.

Primary stage:

Collection of raw data

Secondary stages:

Conversion to interpretable scores

Final stages:

Production of aggregate data

Processes Hardware and firmware collect and

process high-frequency acceleration

signal

Proprietary or custom software

applies algorithms to convert raw

signal to meaningful data

Proprietary or custom software

applies decision rules to perform data

reduction tasks

Outcomes Movement in 1 to 3 axes, usually

expressed in g

Activity counts, energy

expenditure, body position

Total volume (time)

Number of bouts

Length of bouts

Number of breaks

Sources of error Sampling frequency

Non-compliance

Device placement

Lack of transparency in proprietary

algorithms used

Incorrect algorithms used (e.g., for

wrong population)

Lack of validity evidence for

algorithms

Choice of criteria applied for

decisions such as:

- Adjustment for non-wear

- Making “valid day” decisions

- Minimum required days for

reliable and valid data

- Bout length restrictions

- Energy expenditure cutpoints for

sedentary threshold

Page 32 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly

Figure 1. Time spent in different physical behaviors.

Page 33 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly

Please proofread your manuscript carefully. For example:

p. 20

Use a comma to separate the authors:

“. . . 24-hr cycle (Ram, Seirawan, Kumar[,] & Clark, 2010).

[Changed]

Avoid using the same term back-to-back:

“These proportions are approximated in Figure 1. From Figure 1, it is evident . . .”

You may them into one sentence: “These proportions are approximated in Figure 1, which is

evident . . .”

[Modified. It now reads “… approximated in Figure 1, from which it is evident that…” – we feel this

alleviates the concern about the term “Figure 1” being directly back-to-back, and we added the words

“from … it”, in order to make grammatical sense).]

Use the “author, year” format:

“This was demonstrated by Maher, Olds, Mire, and Katzmarzyk (2014), who found that . . .”

[Changed]

Proofread the entire manuscript to make sure when to use “i.e.,” or “e.g.,”

“i.e.,” means “that is”. For example:

It is $10 if you order the ticket online; however, you need to pay double at the door (i.e., $20).

[Checked the entire manuscript as suggested.]

This latter conundrum has been addressed recently from both a measurement perspective and a

behavior modification perspective ([e.g.,] that to reduce sedentary time, one should target

increasing light intensity physical activity).

[This has not been changed to e.g. The author of this section (DR) fully intended to use i.e., (“That is”) –

the detail provided explains “the” exact situation to which I was referring (not an example), specifically it

expands/explains the statement that behavioral interventions designed to disrupt prolonged sitting time

tend to do so by introducing light physical activity such as getting up and walking to the water cooler. I

have added some words to the sentence to remove any ambiguity about what I was trying to say here.

The section now reads “This latter conundrum has been addressed recently from both a measurement

perspective and a behavior modification perspective (i.e., in order to reduce or break up prolonged

sedentary time, behavioral interventions often target increasing light intensity physical activity).]

“. . . when adjusted for total physical activity (i.e., light, moderate and vigorous

intensity????)”

[Reworded to make this clearer: “when adjusted for total physical activity (i.e., the sum of light,

moderate and vigorous intensity physical activity), …”]

Page 34 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly

References

Make sure all the references are in APA format (e.g., use upper case after a colon, use sentence

case for a title, use a comma to separate the journal, volume, and pages, use no period at the end

of the “doi:xxxxx” and so on. Here are just a few examples:

Campbell, D. T., & Stanley, J. C. (1966). Experimental and quasi-experimental designs for

research. Chicago, IL: Rand McNally.

[This has not been changed. Chicago is one of a list of cities that APA says should not be followed by the

state because they are well-known – we refer to APA 5th

edition, p. 217. However, we have italicized the

book title, according to APA guidelines.]

Chastin, S. F. M., Schwarz, U., & Skelton, D. A. (2013). Development of a consensus taxonomy

of sedentary behaviors (SIT): Report of delphi round 1. PloS ONE[,] 8(12)[,] e82313.


[Changed; also, Delphi has been capitalized in the manuscript, as this is a proper noun (named after the

city of Delphi) and proper nouns should be capitalized in book titles and article titles.]

Healy, G. N., Dunstan, D. W., Salmon, J., Cerin, E., Shaw, J. E., Zimmet, P. Z. , . . . Owen, N.

(2008). Breaks in sedentary time: Beneficial associations with metabolic risk. Diabetes

Care, 31(4)[,] 661–666. doi:10.2337/dc07-2046 [PMID:18252901???]

[Changed]

Kim, Y., Lee, J., Peters, B. P., Gaesser, G. A., & Welk, G. J. (2014). Examination of different

accelerometer cut-points for assessing sedentary behaviors in children. PLoS ONE[,] 9(4)[,]

e90630. doi:10.1371/journal.pone.0090630

[Changed]

Kim, Y., Welk, G. J., Braun, S. I., & Kang, M. (2015). Extracting objective estimates of

sedentary behavior from accelerometer data: Measurement considerations for surveillance

and research applications. PLoS ONE[,] 10(2)[,] e0118078.


[Changed]

Maher, C. A., Mire, E., Harrington, D. M., Staiano, A. E., & Katzmarzyk, P. T. (2013). The

independent and combined associations of physical activity and sedentary behavior with

obesity in adults: NHANES 2003‐06. Obesity, 21[,] E730-737. doi:10.1002/oby.20430

[Changed]

Maher, C. A., Olds, T., Mire, E., & Katzmarzyk, P. T. (2014). Reconsidering the sedentary

behavior paradigm. PloS ONE[,] 9(1)[,] e86403. doi:10.1371/journal.pone.0086403

[Changed]

Page 35 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

For Peer Review O

nly

US Physical Activity Guidelines Committee[.] (2008). Physical activity guidelines advisory

committee report, 2008. Washington: DHSS. Retrieved from????????????

[This has not been changed. The citation information is exactly as suggested by the publisher of this

document (i.e., no Web address). The document also is not a Web page (i.e., not changeable, and so

retrieval date is not relevant. Also, The Physical Activity Guidelines Advisory Committee is a proper noun

(title) and therefore should be capitalized even when formatted in sentence case.]

Page 36 of 36



123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960

Documents

University of Strathclyde - Issues and Challenges in Sedentary … · 2020. 9. 30. · Total time spent in sedentary behavior (T) can be derived by summing time in sedentary bouts,